Method for detecting abnormal cases in small samples in multi-dimensional quantization mode

A small-sample, case-based technology, applied in the fields of medical care, medical insurance and software development, can solve problems such as precarious security of medical insurance funds, poor accessibility of supervision, and insufficient service supply capacity

CN105184089AActive Publication Date: 2015-12-23上海金仕达卫宁软件科技有限公司
8 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Current Assignee / Owner
Publication Date
2015-12-23

Smart Images

  • Figure 1
    Figure 1
  • Figure 2
    Figure 2
  • Figure 3
    Figure 3
Patent Text Reader

Abstract

The invention discloses a method for detecting abnormal cases in small samples in a multi-dimensional quantization mode. The method includes the steps of determining disease categories and related items of the disease categories, calculating the weights of the related items of each disease category, optimizing a genetic algorithm for gene selection according to the weights of the related items of the disease categories, determining highly-related item sets of the disease categories, recording moderate degree function results of the disease categories, and screening out disease categories with suitable moderate degree function results and cases of the disease categories by setting a threshold value; determining the similarity of the treatment processes of the disease categories and the abnormal coefficients of the related items of the disease categories; outputting an abnormal coefficient graph, wherein the abnormal coefficient graph includes the abnormal coefficients of the cases and the abnormal coefficients of all indexes related to the cases.
Need to check novelty before this filing date? Find Prior Art

Description

technical field

[0001] The present invention is a multi-dimensional quantitative detection method for abnormal cases in small samples, which involves the fields of medical treatment, medical insurance, software development and the like. Background technique

[0002] With the continuous expansion of the scale of insurance participation in my country, the rapid increase of designated medical institutions for medical insurance, and the rapid increase in the number of insured people, the ability and level of medical insurance protection needs to be continuously improved. This medical insurance management has brought great challenges. The medical insurance agency services in various places are overloaded, the service supply capacity is seriously insufficient, and the accessibility of supervision is poor. Excessive diagnosis and treatment behaviors in designated medical institutions are common.

[0003] In September 2014, the Ministry of Human Resources and Social Security issued ...

Examples

Embodiment

[0048] 1. Determine the related items of the disease

[0049] There is a certain difference between the weight of the disease item and the frequency or frequency of the item. Taking "sodium chloride" in the cataract (H26.9) patient group as an example, in a large number of medical visit data, according to "cataract (H26.9)" The 300 pieces of sample data selected for this diagnosis, and the number of occurrences of the keyword "sodium chloride" are counted (243 times), then the frequency ((TermFrequency)) of "sodium chloride" under this disease type is 0.81. In a nutshell, if a piece of medical treatment data contains keywords w1, w2, ..., wN, their word frequencies in the specific disease-specific medical data are: TF1, TF2, ..., TFN. (TF: termfrequency). Then, the overall weight of the visit data is (TF1+TF2+...+TFN) / N. But there are some problems with this algorithm. In the above example, "sodium chloride" accounts for the total word frequency more than 80% of the total, an...